Skip site navigation (1)Skip section navigation (2)
Date:      Mon, 19 Aug 1996 01:00:24 -0700
From:      "Michael L. VanLoon -- HeadCandy.com" <michaelv@HeadCandy.com>
To:        Bruce Evans <bde@zeta.org.au>
Cc:        msmith@atrad.adelaide.edu.au, nate@mt.sri.com, freebsd-hackers@freefall.freebsd.org, ponds!rivers@dg-rtp.dg.com, ponds!rivers@freefall.freebsd.org, sag.space.lockheed.com!handy@dg-rtp.dg.com
Subject:   Re: sio issues (silo overflows on a pentium, locked in ttywait, etc...) 
Message-ID:  <199608190800.BAA00420@MindBender.HeadCandy.com>
In-Reply-To: Your message of Thu, 15 Aug 96 07:13:54 %2B1000. <199608142113.HAA04204@godzilla.zeta.org.au> 

next in thread | previous in thread | raw e-mail | index | archive | help

>>see the overflows affecting me much.  The funny thing is I don't see
>>overflows on the SLIP server box which is also running 16550A UARTS and
>>is running multiple sessions, but it's not running X and never touches
>>the hard disk.  (Basically it's the same hardware on both boxes, 486/66
>>with 16MB, although my box has an Adaptec 1542B and the server box is
>>IDE).

>Upgrade everything to IDE :-).  Some busmastering SCSI controllers hog
>the bus for too long (e.g., Ultrastor U34Fs sometimes hog it for 160 us
>so they are unusable with single 8250s at speeds > 57600, unusable with

Good point, I guess, but slightly off target.  The problem is not the
SCSI controller, per se, but rather the bus.

There is no bus-mastering "standard" for the ISA bus.  A bus-mastering
controller basically has to guess how long is long enough to stay on
the bus, and there is nothing that enforces that it doesn't stay too
long.  This is one of several reasons not to buy an ISA controller if
you have a better alternative.  VLB is just a faster version of ISA,
so it suffers from the same problem (but at least isn't so dog slow).

EISA and PCI have very well documented standards for bus-mastering.
There are signals that control how long each master is allowed to hold
the bus, and so forth.  You won't see any of this starvation on an
EISA or PCI system because they work in a much better behaved fashion.
Not to mention that they're *several* times faster.

Which means that you *need* to get into the modern era of computing
and throw out that ISA hardware!  If you don't currently have anything
but ISA motherboards, and you want to do it on the cheap, you can do
as little as getting a new 486 motherboard (for way under $100), that
has a PCI bus on it.  You can plug all your old hardware into it and
get a decent PCI SCSI controller.

If you're going to spend money, however, IMHO the current point of
maximum power/dollar is right around the Pentium 100 or Cyrix 6x86
P133+.  A lot of power for not too much money.

-----------------------------------------------------------------------------
  Michael L. VanLoon                                 michaelv@HeadCandy.com
        --<  Free your mind and your machine -- NetBSD free un*x  >--
    NetBSD working ports: 386+PC, Mac 68k, Amiga, Atari 68k, HP300, Sun3,
        Sun4/4c/4m, DEC MIPS, DEC Alpha, PC532, VAX, MVME68k, arm32...
    NetBSD ports in progress: PICA, others...

   Roll your own Internet access -- Seattle People's Internet cooperative.
                  If you're in the Seattle area, ask me how.
-----------------------------------------------------------------------------



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?199608190800.BAA00420>